Rawls's challenge to positivism

In his dissertation, Rawls offers a methodology to discover and validate ethical principles which will function as ‘adequate and justifiable canons for the solution of moral conflicts’, as well as ‘serve as a means for the reform and improvement of common morality’.Footnote ¹⁶ Developing such a normative approach in ethics was no easy task in the prevailing academic atmosphere, however. The dominant positivist philosophy was far from friendly towards such normative projects, accepting merely linguistic analysis of ethical and political terminology. Rawls explains the popularity of the positivist position on ethics by the combination of two factors: the lack of a suitable methodology in ethics, and ethical relativism supported by the interpretations of current anthropological research.Footnote ¹⁷

Thus the first step Rawls sees fit to adopt in his dissertation is to refute the authoritarian and positivist emotivist stands on ethics, both of which regard moral reasoning as futile. His decisive move is to develop a new methodological conception of ethics designed not to be easily rejected by his opponents.Footnote ¹⁸ Rawls's rhetorical strategy consists of two apparently irrefutable moves, the first arguing for the necessity of rational moral reasoning, and the second for its possibility, the latter comprising the main project of the dissertation. The necessity of well-established ethical principles for Rawls follows from the democratic conception of government, according to which laws founded on reasoned public discussion constitute the primary source of political authority. This implies that the modes of rational foundation of ethical principles should form a part of democratic theory, as well as ethical philosophy.Footnote ¹⁹

Since Rawls's strategy in the dissertation is to construct a scientistic methodology for ethics in simulation of the prevailing understanding of the natural sciences along the lines of the new inductive logic, this seems to undermine the claim that Goodman's comments on inductive logic in his 1955 work could have been a major methodological inspiration for Rawls.Footnote ²⁰ Speaking about ‘ordinary methods’, Rawls removes the positivist demarcation between science and non-science by expanding the positivist unity of science principle to include ethics as well. Analogous to natural science, the role of ethical principles is to capture the invariant in actual ethical decisions, and in the case of several sets of principles, the criteria of simplicity and elegance are to guide the choice, just as is often suggested for natural science.Footnote ²¹ The only major modification required to make the IL methodology applicable to ethics, for Rawls, involves a new reading of ‘data’. Instead of observations obtained by scientists, ethical data consist of rational judgements (considered judgements, following Rawls's later terminology) on ethical matters made by reasonable men (competent judges, also following Rawls's later terminology).Footnote ²² The criteria for the competent judges, as well as those for the considered judgements, thus form the main issues in his endeavour to establish an ethical methodology acceptable to the positivistically minded philosophers.

In our explication of Rawls's scientistically oriented IL methodology, we shall rely both on his dissertation ‘Study in the Grounds of Ethical Knowledge: Considered with Reference to Judgments on the Moral of Worth of Character’, and on the essay ‘Outline of a Decision Procedure for Ethics’. As suggested by the opening words of the ‘Outline’ essay, Rawls aims to tackle two questions simultaneously:

The first challenge is to find a reasonable decision procedure in situations of conflict of interest in ethical matters, and the second is to establish the reasonableness of the methodology itself. The basically foundationalist nature of Rawls's methodology becomes clear from the following summary of its three main steps:

The challenge of showing the reasonableness of the IL methodology itself will be met by demonstrating that it serves its purpose of supporting ethical decision-making by yielding ethical principles which cover good ethical decisions already made, and to be made in situations of conflict of interest. Next we shall present Rawls's IL methodology in some detail, and briefly assess the extent to which these two goals are achieved in the dissertation and the ‘Outline’.

Considered judgements and competent judges

The first phase in the Rawlsian IL is to gather the relevant ‘data’ for constructing the ethical principles (IL 1). Rawls's notion of ethical ‘data’ clearly differs from simple observations typically construed as the knowledge basis of science by empiricists, since the ‘data’ is said to consist of actual decisions made in situations of conflict of interest.Footnote ²⁵ Such decisions are characterized as follows: ‘since A, B, C, . . ., and M, N, O, . . . are the facts of the case and the interests in conflict, M is to be given preference over N, O, . . .’.Footnote ²⁶ The decision suggests simply, without compromise, whose preferences are to be preferred over those of others. To avoid circularity, choice is to be made intuitively without relying on any ethical principles, and one is to feel certain about it.Footnote ²⁷

To form a sound basis for ethical theorizing, the decision needs to be made under favourable conditions by a normal, intelligent person meeting certain further requirements to make him a ‘competent judge’. Rawls characterizes the competent judge as one who is knowledgeable about the relevant features of human action in general, such as the likely consequences of an action. He should also be knowledgeable about human interests in a sympathetic manner, and capable of using his imagination in cases where he lacks experience. In order to make a fair and informed choice, the judge has to inquire into all the relevant facts of the case, and allow each party a fair opportunity to state his case. In addition to possessing at least normal intelligence, a competent judge must be willing to use the criteria of inductive logic to determine with an open mind what to believe and what not. To guarantee neutrality, he should be aware of the possible influences of his own emotional, intellectual and moral predilections.Footnote ²⁸

Instead of being one of the parties to the conflict, the competent judge is to be an impartial third party with no gain or loss to expect from his decision, whatever it may be. A considered judgement is further required to be stable in the sense that one can find several competent judges who have rendered an identical decision in a similar situation at other times and places. In other words, the judgements relevant to the construction of ethical principles are such that they ‘are made from day to day on the moral issues which continually arise’.Footnote ²⁹

These features of the competent judge can be assessed independently of the kind of judgements he offers, which guarantees a kind of objectivity for the evidence base of the moral principles in a non-circular manner.Footnote ³⁰ Rawls calls this conception of moral data ‘logical’ or ‘methodological physicalism’, by which he means ‘a principle which embodies an essential rule of scientific methodology, namely, the insistence that the theories and principles of a science be established or refuted wherever possible by objective data which can be checked by the community of investigators, together or individually’.Footnote ³¹

Explication of ethical principles

Having gathered a sufficient set of considered judgements about a number of different conflict situations, the next step is to formulate, explicate in Rawls's terminology, general moral principles to capture what is invariant in the considered judgements (IL 2).Footnote ³² Even though Rawls speaks about explication as a heuristic device, he says nothing about the intermediate steps to be taken between considered judgements and principles.Footnote ³³ The examples meant to exemplify the working of IL in both the dissertation and the ‘Outlines’ offer no help in this respect, since they merely illustrate the justification of ready principles. This accords with his statement that ‘the norms of inductive logic are for the purpose of evaluating the truth of theories once they are formulated. They are not rules for discovery, but canons of proof.’Footnote ³⁴ This is, however, not in harmony with several claims stating that both discovery and justification are equally crucial elements of IL.Footnote ³⁵ In Rawls's examples intended to illustrate his IL, we shall see that its second phase, explication of the principles on the basis of considered judgements (IL 2), plays a minor role after all.

Justification of ethical principles

The task of the moral principles with respect to considered judgements resembles the covering law model of explanation, although Rawls points out that the principles are not intended to offer the causes of the considered judgements.Footnote ³⁶ Likewise, an explication is required to be comprehensive so that it covers ‘all considered judgements, and it is expected to do this with the greatest possible simplicity and elegance’.Footnote ³⁷ This is the chief form of justifying an explication: ‘Like any theory, an explication is tested by the criterion of comprehensiveness.’Footnote ³⁸ The set of principles, if it exists, will be useful in practical life by helping to assess what interests to prefer in situations of ethical conflict.Footnote ³⁹ This means that applying the principles to new cases would yield judgements identical to those made intuitively by the competent judges.Footnote ⁴⁰

Rawls's strong emphasis on comprehensiveness as the criterion of acceptable ethical principles leads us to classify his IL as basically a foundationalist epistemology with the considered judgements as the immediately justified, logically consistent and irrefutable basis, and the moral principles as the mediately justified beliefs.Footnote ⁴¹ The Rawlsian IL is not, however, a purely foundationalist epistemology, since he points out three further situations one may face when testing the success of an explication. The first involves conflicting judgements among different judges, the second an anomaly (a judgement apparently not covered by the principles) and the third a judgement in conflict with some principle. The principles are shown to be reasonable where a solution to such problems can be brought about ‘which, after criticism and discussion, seems to be acceptable to all, or nearly all, competent judges, and to conform to their intuitive notion of a reasonable decision’.Footnote ⁴² The third case allows a fourth form of justification, since, when a principle survives the conflict with a considered judgement and the latter is modified, it is desirable to offer the reason for the mistake in the considered judgement.Footnote ⁴³

The capacity to resolve anomalies and conflicts among considered judgements, and conflicts between a principle and a considered judgement is, for Rawls, a powerful guarantee of the reasonableness of the principles established. Such forms of justification constitute non-foundationalist, coherence-theoretical elements in his IL, since not all considered judgements are regarded as infallible while some may be modified by the principles which again are supported by the modified judgements. The foundationalist element, in contrast, consists of the idea that the considered judgements form a firm knowledge basis on which to explicate general ethical principles, which are supported inductively by these same considered judgements as well as new ones. The coherentist element involves the possibility of correcting a subclass of considered judgements on the basis of the principles explicated, thus allowing a relation of mutual support between the modified considered judgements and the ethical principles explicated.Footnote ⁴⁴ As the emphasis on comprehensiveness clearly indicates, the foundationalist element is for Rawls the main one while the coherentist perspective plays a minor role in IL. In practical applications coherence considerations could, in principle, turn out central in case the considered judgements were mostly inconsistent. As we shall show there is a drastic change in this respect in Rawls's examples in the dissertation and ‘Outline’ essay as compared with the ‘Two Concepts’ essay.

Rawls offers four distinct arguments for the reasonableness of the IL methodology, i.e. to support the view that IL is a reasonable way to establish principles to support ethical decision-making in situations of conflict of interest. Because of its strong foundationalist emphasis IL presupposes a privileged position among moral judgements for the considered judgements. This is what they have, according to Rawls, since the considered judgements are made in favourable conditions to minimize the influence of personal predilections, and their neutrality is further guaranteed by the requirement that several persons have made the same judgement.Footnote ⁴⁵ Rawls's second argument for the reasonableness of IL is that the way of justifying IL resembles the justification of inductive logic, the soundness of which is established by showing that it yields the modes of reasoning applied by scientists.Footnote ⁴⁶ Third, such an IL methodology Rawls claims to be superior to various traditional forms of justifying ethical principles, such as reliance on ethical intuition, the authority of reason, divine revelation or analysing the meaning of ethical terms.Footnote ⁴⁷ The fourth argument for the reasonableness of IL is supposed to be given by demonstrating that the methodology actually works in practice. We shall next take a look at some typical examples Rawls offers as illustrations of the IL methodology in order to assess how far they accord with his portrayal of the three stages of IL.

Assessment of intended illustrations of IL in the dissertation and ‘Outline’

In Part II of the dissertation, Rawls claims to offer a host of examples as illustration of the IL methodology. One of his aims is said to ‘explicate the rational judgements of reasonable men so far as they are applied to the moral character of an agent’.Footnote ⁴⁸ A closer look at the examples reveals, however, that instead of taking a set of considered judgements in situations of conflict of interests as starting points for explication, as suggested by the first phase of IL (IL 1), the treatment begins with ready principles, presuming that these are already available without a process of discovery. This implies that the intended illustrations begin with the third and final step (IL 3), and fail to exemplify the first two steps (IL 1) and (IL 2).

To elucidate, one of Rawls's principles concerning moral character reads as follows: ‘The character of an agent who merely contemplates the doing of an evil action, but does not do it, is not to be judged as bad as the character of an agent who not only contemplates it, but does it.’ Instead of relying on considered judgements as required in (IL 3), the justification of this and the other four principles dealt with relies partly on two principles of practical reason pertaining to the goal-directness of intentional action,Footnote ⁴⁹ and partly on some general common-sense conceptions,Footnote ⁵⁰ as well as teachings from the New Testament.Footnote ⁵¹

Rawls's discussion of another principle that concerns one who merely contemplates the doing of a right action, but does not do it, ends likewise with the remark that it is justified by the principles of practical reason.Footnote ⁵² The principles of practical reason utilized as evidence are:

The treatment of the principles of justice in the ‘Outline’ essay have a similar structure. At the outset we are given seven principles with several sub-principles, such as, ‘each claim in a set of conflicting claims shall be evaluated by the same principles’, and ‘given a group of competing claims, as many as possible shall be satisfied, so far as the satisfaction of them is consistent with other principles’. To illustrate their justification, a few of them are shown to be in harmony with some general conceptions, such as freedom of speech and thought, and the rejection of the institution of inquisition.Footnote ⁵⁴ Here again the process of discovery, i.e. steps (IL 1) and (IL 2), are left out and the evidence applied in the justification consists of general principles instead of decisions resembling the account of the considered judgements.

In summary of our assessment of Rawls's intended illustrations of IL both in the dissertation and the ‘Outlines’ we claim that he fails to demonstrate that the IL methodology works in actual research practice to guarantee sound principles for ethical decision-making. Our analysis of the illustrations reveals a lack of the phase of collecting the considered judgements (IL 1), as well as the discovery phase (IL 2), even though both discovery and justification are purported to be equally crucial elements of IL. Instead of presenting all three steps of IL, Rawls moves directly to ready ethical principles, merely exemplifying their justification (IL 3). Thus, instead of illustrating the inductivist methodology, Rawls's examples both in the dissertation and in the ‘Outline’ essay can be read as illustrations of the hypothetico-deductive model. His hypothetico-deductive model is not, however, the standard empiricist one, since the justification of general principles relies on other general principles and general common-sense conceptions. The considered judgements in the sense of intuitive decisions in situations of ethical conflict of interest do not have a role in the justification process, either. No matter how reasonable IL is in theory, Rawls's main argument for its reasonableness collapses, as he fails to demonstrate that IL works in practice to generate ethical principles to guide decision-making in situations of conflict of interest.

To show that this should not be read as implying that an inductivist approach along the lines of the Rawlsian IL has no relevance to ethical theorizing, we shall briefly consider Aristotle's argument for the relevance of the mean in Nicomachean Ethics II.6 that includes an inductive generalization. The first step in the argument relies on expert opinion, e.g. trainer's decisions concerning the amount of nourishment given to athletes, runners and wrestlers, which should neither be too little, nor too much, but right for the sportsman in question. This evidence, somewhat comparable to Rawls's considered judgements made by competent judges, supports the following generalization: ‘a master of any art avoids excess and defect, but seeks the intermediate and chooses the this-intermediate not in the object but relatively to us’.Footnote ⁵⁵ Next, Aristotle argues that, likewise, a virtuous person strikes a mean in his actions which is not an arithmetical mean, but one based on contextual judgement.Footnote ⁵⁶ This principle is thus gained by generalizing on the basis of consistent conceptions, a move in harmony with the IL methodology.

We may conclude this section on Rawls's inductive logic methodology by suggesting that even though Rawls's examples fail to display how his IL functions in practice, there is no reason to reject totally the inductivist approach in ethical theorizing. Rawls's own examples indicate that the inductivist approach should be complemented by the hypothetico-deductive model. The purpose of the rest of the article is to demonstrate that ethical theorizing is in need of stronger methodological devices than the inductivist and the hypothetico-deductive approaches taken together. The basic reason for this is fairly obvious: the ‘data’ the ethical theoretician is to work with seldom forms a consistent set, but often includes mutually inconsistent beliefs. The Aristotelian saving the appearances and the Rawlsian reflective equilibrium prove their strength exactly in cases like that.

The four phases of SA

The following quotation from Nicomachean Ethics VII.1 is Aristotle's famous passage describing his understanding of the research process when facing problems concerning how to specify particular theoretical notions in a systematic step-by-step approach. Even though the passage is also quoted by Rawls in his dissertation,Footnote ⁵⁹ he makes no further use of it, apparently because its peculiar details were brought to light only somewhat later. Here the problems concern weakness of will (akrasia).

The passage reveals four stages in a philosophical research process:

Here again the aim of the research process is to construct an acceptable, coherent set of the given beliefs as in Daniels's description of the process leading to reflective equilibrium. In order to demonstrate that and how phases (SA 2) and (SA 3) may offer heuristic advice for working back and forth through the given appearances and end up in a reflective equilibrium we need to take a detailed look at each phase of Aristotle's saving the appearances.

Collecting ethical ‘data’ in SA

First we need to show that Aristotle's ‘appearances’ cover Daniels's starting points of ethical research, i.e. include ‘considered judgments of particular instances or cases . . ., principles or rules we believe govern the considered judgments, and theoretical considerations that we believe bear on accepting both the considered judgments and the principles’. The term ‘appearance’ expresses how matters appear to us, and it includes not only how they do so in observation and experience, or in intuitive judgements, but also in theoretical reasoning.Footnote ⁶¹ Saving appearances relevant to ethical issues thus means saving current views on ethical matters at various levels of abstractness.

The term ‘appearance’ is ambiguous in an interesting way that reveals a major difference between SA and the foundationalist emphasis in Rawls's IL. Things may appear to us in a misleading way, but also reveal what is truly the case, as exemplified by over-generalizations reflecting the particular social position of a group of people.Footnote ⁶² For instance, the highest goal of human life appears to be different to people in different circumstances, like a politician, one seeking wealth or pleasure, or a healthy and a sick person.Footnote ⁶³

Another epistemic notion characterizing the basis and criterion of philosophical inquiry in Aristotle is ‘reputable opinions’ (endoxa), meaning the opinions of reputable men (endoxoi).Footnote ⁶⁴ Excluding children and madmen, Aristotle's classification of reputable opinions in his textbook on dialectic, the Topics, embraces the opinions of all, the majority, the wise or the most authoritative of the wise, as well as experts in various fields.Footnote ⁶⁵ The term ‘reputable’ expresses the social aspect of the knowledge basis, since it is social recognition that lends credibility to the beliefs of reputable people, and makes them worthy of serious consideration. Even though their connotations differ, in ethics the extensions of the terms ‘appearances’ and ‘reputable opinions’ seem to coincide.Footnote ⁶⁶

This shows that the starting points for ethical theorizing in the Aristotelian SA include considered judgements with their fairly strict requirements imposed by the Rawlsian IL, but they also embrace the other elements Daniels includes in the starting points for building a reflective equilibrium, i.e. ‘principles or rules we believe govern the considered judgements, and theoretical considerations that we believe bear on accepting both the considered judgements and the principles’.

Critical scrutiny of the ethical ‘data’ in SA

As Plato's dialogues and Aristotle's Nicomachean Ethics reveal, inconsistencies on ethical matters abound among given views. This is also the background assumption in Daniels's ‘working back and forth’ idea, though not the main focus in Rawls's basically foundationalist IL. It is therefore most important to have some methodological ideas about how to work with inconsistencies to distinguish the true from the false. A great merit of SA is its heuristics for explicating and solving inconsistencies (SA 2). In this, Aristotle builds on the logical tools developed by his teacher and, like Plato's Socrates, applies the elenchus, although without an explicit question-and-answer structure, to argue conflicting views both for and against to detect their strengths and weaknesses.Footnote ⁶⁷ Critical assessment also involves conceptual clarification and formulation of the given views in a clearer way to gain a better grasp of their implications. This is the first step in our attempt to clarify further Daniels's description of the research process leading to reflective equilibrium.

Generating the solution in SA

A typical feature of the Aristotelian SA is that the phases of critical scrutiny (SA 2) and constructing the solution (SA 3) are often closely related. Since the critical scrutiny aims to disclose what is true and what is false in the appearances, it simultaneously gathers information for the construction of the solution to the given problem. Sometimes the solution can be produced simply by combining what is left standing of the appearances after the critical phase, but this is obviously not always the case as innovative steps may be needed to construct an umbrella conception to ‘save the appearances’. The simple case is engendered by two or more conflicting over-generalizations, since a successful argument for and against already reveals how to qualify each to create a synthesis.

Plato's famous method of collection and division in the Phaedrus is an early example of building a synthesis in the case of conflicting views.Footnote ⁶⁸ The dialogue exemplifies this through Socrates’ two speeches about love, the first one based on the hypothesis that love is a mad desire for the physical pleasure of beauty, and the other defining love as divine madness. The clue to the solution is the specification of an appropriate generic term (collection phase), here madness, and then dividing it into proper subclasses.Footnote ⁶⁹ A somewhat similar technique is applied by Aristotle in his treatment of weakness of will in Nicomachean Ethics VII.3, where the challenge is posed by Socrates’ claim, against the majority view, that one cannot act against one's knowledge. Aristotle's solution, the details of which are much disputed by scholars, is based on several qualifications of the concept of knowledge.Footnote ⁷⁰ To solve more demanding problems involves, in contrast, genuine moments of invention with no mechanical rules of discovery. However, the critical phase may offer significant help in determining where to seek the solution.Footnote ⁷¹ As we shall see, Rawls's treatment of the justification of punishment in ‘Two Concepts’ falls within the simple types of problem arising from conflicting over-generalizations. In developing one's capacity to solve theoretical problems of various kinds to generate reflective equilibriums, one may profit greatly by studying suitable literature, for instance Aristotle's treatises. This is the second move to specify Daniels's notion of research process leading to reflective equilibrium.

Justification in SA

The fourth and final phase in SA consists of the justification of the solution (SA 4) which, according to the excerpt quoted above, presupposes establishing that the problem is solved and that the appearances are saved. This also clarifies what, in fact, is involved in a reflective equilibrium of given conceptions, and its justification. Aristotle's requirement that the problems are solved clearly demands the basic condition of coherence, i.e. that the solution is consistent. The second justificatory task, responsible for the title saving the appearances, is to demonstrate that the initially conflicting appearances, or most of them, are included in the solution in a suitably revised form. In the case of qualified over-generalizations, this causes no particular difficulty, as already exposed with the help of the example from Plato's Phaedrus.

A further form of justification pointed out and sometimes applied by Aristotle, although not mentioned in the above passage, corresponds to Rawls's IL suggestion that, when considered judgements need a modification, the reason for the anomaly should be given.Footnote ⁷² Aristotle illuminates the importance of explaining why a false view arose in the first place as follows:

‘Saving the appearances’ thus appears to lead to the kind of end result of research characterized by ‘reflective equilibrium’: the prior beliefs, or at least most of them when suitably modified, need to find their place in the end result. We hope that our presentation of the SA methodology has by now revealed that the four Aristotelian moves (SA 1)–(SA 4) offer a stronger heuristics for generating a reflective equilibrium than Daniels's three-step conception (D 1)–(D 3). For instance, in addition to generating hints for solving the problem in question (SA 3), the argumentative procedure in the critical scrutiny phase (SA 2) also yields material for justifying the solution in the manner according to (SA 4), and as required by coherence theory. We need to point out, however, that because of the research heuristics, SA is not to be identified merely as a form of coherence theory, since coherence theory in the standard sense is merely an epistemological theory of justification without elements of theory generation.

Rawls's IL compared with Aristotle's SA

The Aristotelian SA leading to a reflective equilibrium deviates fundamentally from the Rawlsian IL. The differences concern every phase of the research process, the understanding of what kind of ‘data’ to adopt as the starting point of ethical theorizing, the nature of problems arising from the ‘data’, the heuristics of generating ethical principles and the justification of the principles. The major cause of the differences is the logical structure of ‘data’, since in IL the ‘data’ are for the most part mutually consistent, whereas Aristotle focuses on conflicts within the ‘data’. The Stagirite allows a much wider variety of ‘data’ as they may consist of various kinds of theoretical views held by almost everyone or by philosophers, while in IL the ethical ‘data’ comprises particular decisions made by competent judges concerning situations of conflict of interest.

It is worth noting that in spite of Rawls's strong scientistic ambitions, his considered judgements involve a strong Aristotelian element. Instead of being positivist-type observations, considered judgements made by competent judges are endoxa in the Aristotelian sense. Considered judgements are, first of all, not observations but judgements and, second, the competent judges are a particular subgroup of reputable people.Footnote ⁷⁴ Aristotle would, however, relax the Rawlsian requirement that the competent judge needs to be a party external to the situation, since he accepts interest-laden decisions, which are to be saved by revealing their underlying interests.Footnote ⁷⁵

The intended illustrations in Rawls's dissertation and the ‘Outline’ essay include even stronger Aristotelian elements. The principles and general beliefs Rawls aims to justify and those he uses as evidence are without doubt Aristotelian endoxa. To explain why he seldom sees a need to begin by collecting considered judgements and explicating new principles, Rawls notes that ‘the important theories of the past should be known, since it is highly possible that some one of them, or some combination of them, may be correct after certain adjustments and changes have been made’.Footnote ⁷⁶ Here he heavily but unconsciously undermines the reasonableness of his own strongly foundationalist IL methodology in favour of the reasonableness of the hypothetico-decuctive one. Interestingly enough, the remark bears close affinities to a comment by Aristotle:Footnote ⁷⁷

Aristotle is evidently expressing his belief in the reasonableness of the saving the appearances methodology.

To account for the differences between IL and SA, we could point out that Aristotle and the early Rawls emphasize contrary aspects in ethical research. While Aristotle highlights conflicts in given appearances and thus has a strong coherentist involvement, Rawls stresses consistency, implying a foundationalist emphasis. When working with more demanding theoretical problems, Rawls is also bound to face inconsistent appearances, as we shall see in the next section. Interestingly enough, his approach then has a strong Aristotelian feel.

Significance of ‘Two Concepts of Rules’

We shall demonstrate in this section that Rawls's defence of utilitarianism in ‘Two Concepts of Rules’ when dealing with the justification of punishment conforms to the Aristotelian principle of saving the appearances. As has been shown in the preceding section the SA methodology offers a more detailed description of how to reach reflective equilibrium than Daniels's ‘working back and forth’ account. However, the essay merits a methodological analysis on its own for its importance among Rawls's works. Even though overshadowed by his main works, its significance is shown by the numerous translations as well as by its enduring position in anthologies dealing with moral philosophy. It is incorporated into Rawls's Collected Papers, and has attracted interest in jurisprudence, sociology, political science, psychology, economics and organization studies. Quite recently the Journal of Classical Sociology published a special issue edited by Anne Warfield Rawls focusing on the theoretical relevance of the ‘Two Concepts’ to contemporary sociology.

As the title of the essay indicates, Rawls's main aim in his ‘Two Concepts’ is to argue for two notions of rules, namely, the summary and the practice conceptions. The distinction relates to two forms of ethical argument, ‘justifying a practice and justifying a particular action falling under it’.Footnote ⁷⁹ By applying the distinction to two cases, the justification of punishment and promise, Rawls aims to defend utilitarianism against some traditional objections by developing a version of it called ‘practice-utilitarianism’,Footnote ⁸⁰ in which the rule-based conception of practice plays an important role, and helps to avoid certain typical objections relating to the justification of the institutions of punishment and promise. Rawls does not, however, aim at a complete defence of the utilitarian moral doctrine, but intends to make ‘a logical point’ which in itself ‘leads to no particular social or political attitude’.Footnote ⁸¹ Thus the ‘Two Concepts’ essay helps us to understand Rawls's later attitude to utilitarianism.Footnote ⁸²

The ‘Two Concepts’ includes a revealing methodological ambiguity, however. Having briefly presented his solution to the justification of punishment (JP), Rawls remarks that the two theories of punishment, the utilitarian and the retributive view, have been reconciled ‘by the time-honored device of making them apply to different situations’.Footnote ⁸³ In the same vein, he states that one of his aims is to formulate utilitarianism ‘in a way which saves it from several traditional objections’,Footnote ⁸⁴ and ‘allows for the sound points of its critics’.Footnote ⁸⁵ These remarks reveal a connection with Plato's method of collection and division, further extended by Aristotle's SA. However, at the beginning of the essay Rawls refers to his IL methodology by pointing out that the aim of the paper is to ‘state utilitarianism in a way which makes it a much better explication of our considered moral judgments than these traditional objections would seem to admit’.Footnote ⁸⁶ For the meaning of ‘explication’, he refers to the ‘Outline’ essay.Footnote ⁸⁷

Because of the obvious discrepancy between the first two remarks and the third, our aim is to go deeper into the methodological practice of the essay by clarifying the structure of the JP argument in particular. We shall demonstrate point by point how Rawls's methodology in solving the conflict between the two theories of punishment in the ‘Two Concepts’ follows the lead of the Aristotelian SA. The notion of ‘data’, the nature of the research problem, and the way of discovery, as well as justification, accord with the SA rather than the IL methodology. This supports our claim that Rawls's ‘Two Concepts’ involves a move towards a new methodological practice, resembling his descriptions of RE. We shall next show how the four phases of SA offer a more satisfactory reading of the JP argument than an IL reading.

Collecting ethical ‘data’ in JP

At the beginning of his discussion of JP in the ‘Two Concepts’ essay, Rawls notes that, in spite of the shared agreement concerning punishment in the sense of attaching penalties to the violation of legal rules, the topic has been a troublesome one, the problem being the moral justification of punishment. He points out that none of the justifications has so far ‘won any sort of general acceptance; no justification is without those who detest it’.Footnote ⁸⁸

Rawls next introduces two competing views on the justification of punishment, the retributive and the utilitarian views, in accordance with the first phase of SA. The retributive view holds that ‘punishment is justified on the grounds that wrongdoing merits punishment’ and the severity of the appropriate punishment should depend on the depravity of the wrongdoing. The utilitarian view, by contrast, looks forward, maintaining that justifiable punishment effectively promotes the interests of the society and that punishment is justifiable only by reference to its probable consequences.Footnote ⁸⁹

We may note immediately that the retributive and utilitarian justifications of punishment do not constitute the kind of knowledge basis for ethical decision-making as required by Rawls's IL. First of all, they are not decisions made spontaneously in moral cases of conflicting interests, but are, instead, general theoretical conceptions with their own justification. This implies that they satisfy the condition of being ‘considered judgements’ made by ‘competent judges’ in the Aristotelian sense of reputable opinions. Rawls himself states that both positions are worth serious consideration since ‘intelligent and sensitive persons have been on both sides of the argument’.Footnote ⁹⁰

Critical scrutiny of the ethical ‘data’ in JP

In carrying out our analysis, it is helpful to note that the phase of critical scrutiny (SA 2) may already involve support for the solution, and no strict order of presentation is presupposed in SA. After a preliminary discussion of the two conceptions of punishment, Rawls proceeds to the solution,Footnote ⁹¹ and then, in order to offer further justification to the solution suggested, continues by exploring some objections to utilitarianism presented from the retributionist point of view.

To start with, Rawls notes that there is a contradiction between the retributive and utilitarian views by pointing out simply that the way each is stated makes ‘one feel the conflict between them’.Footnote ⁹² Having next suggested his solution to the conflict, he focuses on the retributionists’ challenge of ‘whether utilitarianism doesn't justify too much’.Footnote ⁹³ Isn't the utilitarian committed to accepting the punishment of innocent people, for instance, in a situation where society is shocked by a terrorist attack? According to the retributionist, it is justifiable to punish only the real terrorists, while he could argue that a utilitarian might punish justifiably innocent people if that would effectively promote the interests of the society. This could be the case, for instance, when the real terrorists are extremely difficult to catch, and the interests of the society are advanced when some person or a group is charged and punished credibly and publicly for the attack. Isn't it in the interest of society to instil fear in future terrorists and, more importantly, to restore the feeling of security among its people?

In his defence, Rawls argues that such considerations are excluded for the utilitarian on the basis of the concept of punishment: ‘utilitarians agree that punishment is to be inflicted only for the violation of law’.Footnote ⁹⁴ This is thus an assumption shared by both parties. For closer consideration of the retributionist stand, he cites a lengthy passage from Carritt, which begins: ‘the utilitarian must hold that we are justified in inflicting pain always and only to prevent worse pain or bring about greater happiness’.Footnote ⁹⁵

We need not cite the passage in full, since this suffices to expose Rawls's objective. He rejects Carritt's criticism as futile because of the failure to specify the pertinent institution or institutions referred to by the ambiguous ‘we’.Footnote ⁹⁶ This observation prepares the ground for constructing the solution, as sometimes happens at the critical scrutiny phase (SA 2).

Generation of the solution in JP

In accordance with the third phase of SA, Rawls shows that limiting the scope of application of the utilitarian and retributive views on justification reconciles them so that they no longer contradict each other. Following the suggestion to specify the agents and institutions relevant to punishment, he draws the distinction between ‘justifying a practice as a system of rules’ and ‘justifying a particular action which falls under these rules’.Footnote ⁹⁷ This yields the so-called practice-utilitarian view, the proper scope of which is the justification of the social practice of punishment as a system of rules. The retributive view, in contrast, is appropriate with regard to questions about ‘application of particular rules to particular cases’.Footnote ⁹⁸

In a democratic society there are, in fact, two distinct institutional actors associated with punishment, the legislator and the judge. The forward-looking utilitarian view is, hence, the viewpoint of the legislator, whereas the backward-looking retributive view is the viewpoint of the judge.Footnote ⁹⁹ When performing the office of a judge, a person cannot adopt some sort of utilitarian legislator's viewpoint and distribute punishment as a means of serving the interests of society. A judge should strive to achieve a state of affairs in which only the guilty are punished. This way of resolving the conflict by means of specifying a general concept and then dividing it appropriately closely resembles Plato's methodology of collection and division, and thus explains Rawls's statement that he ‘reconciles the two views by the time-honored device of making them apply to different situations’.

Justification of the solution in JP

Finally we shall show how, in his justification of the solution, Rawls adopts the various argument forms typical of SA. Rawls rejects the retributionist's worry that a utilitarianist might accept that punishing the innocent exemplifies justification for utilitarianism drawn at the critical phase (SA 2). Showing that the conflict in the appearances has been resolved is achieved by confining each position to a particular social context, the retributive conception to the judge and the utilitarian to the legislator. This is a way of saving the appearances, as noted by Rawls himself when stating that the solution simultaneously ‘allows for the apparent intent of each side’ and ‘seems to account for what both sides have wanted to say’.Footnote ¹⁰⁰ The retributionists have correctly insisted ‘That no man can be punished unless he is guilty, that is, unless he has broken the law’.Footnote ¹⁰¹ The utilitarians, by contrast, are concerned with the institution of punishment as a system of rules to foster the good of society effectively.Footnote ¹⁰²

Rawls further illustrates the soundness of the solution with the example of a boy putting two distinct explanatory demands to his father. In one, the boy requires the explanation of the imprisonment of a particular person, and in the other the explanation of the institution of punishment. To clarify the difference between these two questions, Rawls formulates them with the help of illuminating contrasts.Footnote ¹⁰³ The first question asks ‘why was J punished rather than someone else?’ and the second asks ‘why do people punish one another rather than, say, always forgive one another?’Footnote ¹⁰⁴

Rawls furthermore applies the argument form included in both IL and SA by suggesting how one can miss the distinction between justifying a practice and justifying a particular action falling under it. This is where the two concepts of rules, the practice and the summary conception, step in. According to the former, rules constitute a practice by defining what actions are appropriate to the practice in question while, according to the latter, rules are pieces of advice to speed up decision-making in particular cases that tend to recur. A major difference between these two kinds of rule is the order of logical priority. While practice-rules are logically prior to particular actions, particular actions are logically prior to the summary rules:

Thus it is logically impossible to perform a particular action outside the particular practices setting the stage for it. For instance, getting a free-kick in a game of soccer is possible only through playing the game in question, i.e. by following the rules which define the game.Footnote ¹⁰⁶

Even though summary rules, or what could also be called ‘strategic rules’, may contain historically tested practical knowledge, they are only guides for decision-making, and there is nothing wrong in questioning their applicability to particular cases. For instance, it is possible to play soccer by adopting strategic rules (strategies) that have proved successful in previous games. These strategic rules do not, however, define soccer as a game, because it is possible to play the game without adopting them.Footnote ¹⁰⁷ Rawls emphasizes, however, that not all rules fit nicely into these two categories, or that only one of these conceptions is the right view.Footnote ¹⁰⁸ According to him, it is the summary conception of rules that fails to perceive the significance of distinguishing justifying a practice as a system of rules and justifying particular actions falling under it, while the practice conception stresses the significance of the distinction.Footnote ¹⁰⁹

Summary of the analysis of JP

We have shown that Rawls's JP argument in the ‘Two Concepts’ accords neatly with the four stages of Aristotle's SA methodology rather than with the three phases of his IL methodology developed in the dissertation and repeated again in the ‘Outline’ essay. Clearly, the starting point of the investigation, the retributive and the utilitarian forms of justifying punishment, are not decisions made by competent judges in situations of moral conflict of interest, as the ‘data’ characterized in IL (IL 1) are theoretical principles worth serious consideration, corresponding to Aristotle's endoxa (SA 1). Since these views are in conflict, the task for the philosopher cannot be simply to explicate a more general conception to cover them (IL 2). SA has an adequate approach to such problem situations by suggesting that one examine both views critically before attempting a solution to the conflict. To extract the truth involved, Rawls proceeds by articulating the strengths and weaknesses of both positions, as if following the Aristotelian guidelines (SA 2). Having constructed a synthesis of them (SA 3), he justifies this by demonstrating that, in addition to having resolved the conflict, the intention of the proponents of both views is saved. Furthermore, he explains the origin of the conflict (SA 4).

Perhaps the most striking divergence between the Aristotelian SA and the Rawlsian IL methodology concerns the heuristics of discovery. While Rawls explicitly speaks about ‘explication’ as a ‘heuristic device’, his characterization of ‘explication’ says very little about how the principles are to be constructed. Resolving conflicts between prevailing theoretical conceptions through their criticism offers a fairly strong heuristics for creating theoretical ethical principles at any rate, if not a ‘logic’ or ‘mechanical methodology’ of discovery.Footnote ¹¹⁰ This concludes our argument that Rawls's actual methodological practice in ‘Two Concepts’ follows the Aristotelian methodology of saving the appearances (SA) rather than his inductive logic methodology (IL) of the dissertation. Our analysis of the argument structure of his manner of solving the problem of justification of punishment illuminates how the SA methodology yields a richer description of discovery and justification than Daniels's (D 1)–(D 3).